Dim efi
- Thread starter KitCarlson
- Start date
-
KitCarlson
Well-Known Member
Del,
I designed and built my own. It uses an Atmel AVR ATmega128 micro controller.
Here is more information: http://pwp.att.net/p/s/community.dll?ep=16&groupid=372867&ck=&ck=
If you go to photos albums, there are pictures of the populated PCB. Also see system details.
I designed and built my own. It uses an Atmel AVR ATmega128 micro controller.
Here is more information: http://pwp.att.net/p/s/community.dll?ep=16&groupid=372867&ck=&ck=
If you go to photos albums, there are pictures of the populated PCB. Also see system details.
magnumdust
Well-Known Member
Out of curiousity, why not use a GM tbi? There is so much aftermarket for those TBI's that you could get drivability and still have good performance.
KitCarlson
Well-Known Member
I could do that, but there are no limitations when I do things from scratch. I have complete control over the hardware and software. When I go to MPI, i will pull the injectors, and open the air stream, to use the TBI as a throttle body. I will also use a better method of idle speed control.
I have a huge amount of software already developed, new applications are a minor amount of work. It is a hobby for me.
I have a huge amount of software already developed, new applications are a minor amount of work. It is a hobby for me.
KitCarlson
Well-Known Member
Del,
If you want to learn, I am here to help. I use CodeVision AVR for writing C on the EMS2, and Basic4Android for the Android applications. I am not good at MicroSoft, I topped out at VB6, so moving on to simpler Android.
There is a free version of CVAVR that will work for small programs, standard package about $120. The B4A was about $70 for 2 years support.
The AVR micro controllers are powerful and easy to use. There are some faster ones out there, but implementing them is not as much fun. Once you understand a few AVR's it is easy to use different members of the family. They have 6 pin versions (=), all the way to 100 pins, and share instruction set and many peripherals. The six pin is the size of the =.
Easy may be relative, I have been working in product development over 30 years. I learn a bit at a time, but have (or had) good retention.
If you want to learn, I am here to help. I use CodeVision AVR for writing C on the EMS2, and Basic4Android for the Android applications. I am not good at MicroSoft, I topped out at VB6, so moving on to simpler Android.
There is a free version of CVAVR that will work for small programs, standard package about $120. The B4A was about $70 for 2 years support.
The AVR micro controllers are powerful and easy to use. There are some faster ones out there, but implementing them is not as much fun. Once you understand a few AVR's it is easy to use different members of the family. They have 6 pin versions (=), all the way to 100 pins, and share instruction set and many peripherals. The six pin is the size of the =.
Easy may be relative, I have been working in product development over 30 years. I learn a bit at a time, but have (or had) good retention.
BillGrissom
Well-Known Member
Kit,
We are lucky to have you in the classic Mopar fold. I also hope to do some custom controllers for my Mopars, and even bought a few kits (TI MSP430 & a PLD), but no time. In my day job, I occasionally write programs for Windows or PLC's.
I also grew up w/ C, QB45, and VB6. I started VB.net several years ago and find it much nicer than VB6. It is more like using VBA in Excel. There are numerous pre-built objects you can call to make your code easy and concise, ex. parsing a text string into an array based on commas is a 1-line operation. It is just hard to find all the functions in the forest. It also never blue screens since your code is actually script that runs thru the CLR interpreter. C# is almost identical, just for those who want to imagine they are programming in C. Most good programmers I know are middle-aged and experienced. I work with several young EE graduates who don't know much programming, other than writing some Matlab m code.
We are lucky to have you in the classic Mopar fold. I also hope to do some custom controllers for my Mopars, and even bought a few kits (TI MSP430 & a PLD), but no time. In my day job, I occasionally write programs for Windows or PLC's.
I also grew up w/ C, QB45, and VB6. I started VB.net several years ago and find it much nicer than VB6. It is more like using VBA in Excel. There are numerous pre-built objects you can call to make your code easy and concise, ex. parsing a text string into an array based on commas is a 1-line operation. It is just hard to find all the functions in the forest. It also never blue screens since your code is actually script that runs thru the CLR interpreter. C# is almost identical, just for those who want to imagine they are programming in C. Most good programmers I know are middle-aged and experienced. I work with several young EE graduates who don't know much programming, other than writing some Matlab m code.
66plyValiant
Well-Known Member
good thread, I just came across this; I've been having pretty good luck with the kits I sell and I ship anywhere in the US or canada
hows your EFI project coming along?
hows your EFI project coming along?
KitCarlson
Well-Known Member
Bill,
Thank you for the kind words. I am lucky that my education, and varied work experience made it possible. I was fortunate in the process, of adding to skill sets in the workplace, when completing assigned projects. It is about non-stop learning.
I looked at the MSP430 about 15 years ago when it first came out. At that time the ADC and timers were not going to be enough to do EFI and ignition.
The ATmega128 fit nicely, it had enough timer channels (8+) to get the job done and sufficient on-chip RAM, EEPROM and FLASH resources. It did not require glue logic. Most engine management duties are about measuring time, and setting time events. I used an ISR based state machine, that was driven by engine timing sensors. The code is very small, the largest part of code is the user interface. The handing of the setting is done in a shared way, for real time tuning without any hesitation.
Some like the new Cortex controllers, however I have not tried them yet.
Thank you for the kind words. I am lucky that my education, and varied work experience made it possible. I was fortunate in the process, of adding to skill sets in the workplace, when completing assigned projects. It is about non-stop learning.
I looked at the MSP430 about 15 years ago when it first came out. At that time the ADC and timers were not going to be enough to do EFI and ignition.
The ATmega128 fit nicely, it had enough timer channels (8+) to get the job done and sufficient on-chip RAM, EEPROM and FLASH resources. It did not require glue logic. Most engine management duties are about measuring time, and setting time events. I used an ISR based state machine, that was driven by engine timing sensors. The code is very small, the largest part of code is the user interface. The handing of the setting is done in a shared way, for real time tuning without any hesitation.
Some like the new Cortex controllers, however I have not tried them yet.
BillGrissom
Well-Known Member
My idea with the MSP430 is a simple "optimizer" to improve my Holley Pro-jection TBI. I want to add manifold pressure (MAP) and inlet air temp (IAT) sensors to the mix, instead of relying solely on throttle position (TPS) and rpm, which seems to make it always quirky. I would still have the Holley controller. Indeed, it is kind of nice to tune by just turning a few knobs, rather than adjust a fuel map on a PC.
For spark timing control, it seems best to use dedicated logic chips to count crank wheel pulses, hence the Programmable Logic Device (PLD). I know that Megasquirt uses CPU code directly to count crank pulses, which keeps it pretty busy. I wonder how OEM engine controllers do it.
For spark timing control, it seems best to use dedicated logic chips to count crank wheel pulses, hence the Programmable Logic Device (PLD). I know that Megasquirt uses CPU code directly to count crank pulses, which keeps it pretty busy. I wonder how OEM engine controllers do it.
KitCarlson
Well-Known Member
Bill,
Back in the 80's when they used 8051 controllers they did use a divider chip in the input in the Motronic systems. The update rate was slow, today's processors are easily 20 to 100' times faster. I do updates every ignition reference signal.
It is not about counting, it is measuring. Here is how it works. A timer automatically captures the time of an ignition reference signal, and a interrupt is generated. The prior time is then subtracted from the present time, that is period between pulses. The subtraction is fast, only a few clock cycles. The other calculations are done in back ground, in the available time. The results of the prior calculations, are used in the timer interrupt, to set the ignition and fuel injection timer compare events. A timer compare event is an interrupt, that can be set to happen, x clock ticks from a given time reference. The x, comes from the calculations, look up tables and other settings.
So, an ignition control needs to energize the coil, release it at the ignition event, stay open for the spark duration, plus the time delay to the start of coil charge. The control times are variable, based on RPM, and MAP. The advance control in timer ticks is inserted in timer compare register, a one line code that again only takes a few CPU cycles. The timing events are like stepping stones. The state machine navigates the path, acting similar to the PLD, that you desire, except it has the ability for software variable input. The soft variables input, or advance control is what it is all about. Fuel timing control is much the same, and and in a few respects is easier. There are many shared calculations, RPM and MAP, so doing both in one unit is efficient. There is also interplay between the ignition and fuel, separate controls can be a significant compromise in optimized tuning.
I was not trying to be negative about your plan, of MSP430 and PLD, I just do not see how it can be be done. The hardest part of design is often the starting point, choosing the parts. I consider myself quite lucky, in my design career of over over 35 years, I always successfully completed projects based on my start plans. While most consider me highly intuitive or a psychic, it more about sorting out the data sheets. Sometimes data sheets are marketing driven, limitations are buried with white lies.
I also start by purchasing some sort of quick start development board, and writing simple code to check the capability before I commit. Tests like that can be done in a day or two, and provide great insurance and comfort.
I will share (give) you the ATmega128 development board I started with back in about 2003. With the addition of a few I/O components on a second card it will manage an engine.
Back in the 80's when they used 8051 controllers they did use a divider chip in the input in the Motronic systems. The update rate was slow, today's processors are easily 20 to 100' times faster. I do updates every ignition reference signal.
It is not about counting, it is measuring. Here is how it works. A timer automatically captures the time of an ignition reference signal, and a interrupt is generated. The prior time is then subtracted from the present time, that is period between pulses. The subtraction is fast, only a few clock cycles. The other calculations are done in back ground, in the available time. The results of the prior calculations, are used in the timer interrupt, to set the ignition and fuel injection timer compare events. A timer compare event is an interrupt, that can be set to happen, x clock ticks from a given time reference. The x, comes from the calculations, look up tables and other settings.
So, an ignition control needs to energize the coil, release it at the ignition event, stay open for the spark duration, plus the time delay to the start of coil charge. The control times are variable, based on RPM, and MAP. The advance control in timer ticks is inserted in timer compare register, a one line code that again only takes a few CPU cycles. The timing events are like stepping stones. The state machine navigates the path, acting similar to the PLD, that you desire, except it has the ability for software variable input. The soft variables input, or advance control is what it is all about. Fuel timing control is much the same, and and in a few respects is easier. There are many shared calculations, RPM and MAP, so doing both in one unit is efficient. There is also interplay between the ignition and fuel, separate controls can be a significant compromise in optimized tuning.
I was not trying to be negative about your plan, of MSP430 and PLD, I just do not see how it can be be done. The hardest part of design is often the starting point, choosing the parts. I consider myself quite lucky, in my design career of over over 35 years, I always successfully completed projects based on my start plans. While most consider me highly intuitive or a psychic, it more about sorting out the data sheets. Sometimes data sheets are marketing driven, limitations are buried with white lies.
I also start by purchasing some sort of quick start development board, and writing simple code to check the capability before I commit. Tests like that can be done in a day or two, and provide great insurance and comfort.
I will share (give) you the ATmega128 development board I started with back in about 2003. With the addition of a few I/O components on a second card it will manage an engine.
KitCarlson
Well-Known Member
Opps, I made double post.
PS. Long ago I evaluated the MSP430. It would work better for a multi-meter design.
PS. Long ago I evaluated the MSP430. It would work better for a multi-meter design.
BillGrissom
Well-Known Member
Thanks, but I already have a spare Pro-jection controller (the big box) that I got on ebay for $15 a few years ago. I have run both and it seems like the smaller "2D" box is more problematic. It seems to overheat easily if it gets any sun on it, then stops pulsing the fuel. Indeed, I suspect the original MSD-labelled boxes are best since they use more rugged Weather-pack connectors. I also have been collecting Commander 950 sets, and now have enough for all my old Mopars. The last one I got for just $250 buy-it-now on ebay, which is much less than Megasquirt kits and a whole lot less effort. The Pro-jection plans are just for hobby. I grew so frustrated using one for 10 years that I figured I can make it work right.
BillGrissom
Well-Known Member
Kit,
The MSP430 would be just for the Pro-jection improvement, not combined with a PLD. I couldn't resist TI's development kit. I recall just $5, selling at a loss to get engineers using them. I also found a book that tutors on using it.
The PLD is also a kit, that I got cheap on ebay. I got it mainly to learn PLD's, but thought it might work for spark timing from a 36-1 wheel that I put on my Dart. If that proves too much, I have a Ford EDIS spark box that can decode the crank pulses. My controller would then just need to delay the Ford box's SAW square wave, which is how most Megasquirters control spark. I wonder if the Holley Commander 950 can already do that since its interface to a GM 8-pin HEI module is very similar for spark timing command.
For now no time to play. I need to finish working on my roof before the rains return to NorCal. I'd be up there now, but it is 101F and sunny.
The MSP430 would be just for the Pro-jection improvement, not combined with a PLD. I couldn't resist TI's development kit. I recall just $5, selling at a loss to get engineers using them. I also found a book that tutors on using it.
The PLD is also a kit, that I got cheap on ebay. I got it mainly to learn PLD's, but thought it might work for spark timing from a 36-1 wheel that I put on my Dart. If that proves too much, I have a Ford EDIS spark box that can decode the crank pulses. My controller would then just need to delay the Ford box's SAW square wave, which is how most Megasquirters control spark. I wonder if the Holley Commander 950 can already do that since its interface to a GM 8-pin HEI module is very similar for spark timing command.
For now no time to play. I need to finish working on my roof before the rains return to NorCal. I'd be up there now, but it is 101F and sunny.
67Dart273
Well-Known Member
The Commander 950 can indeed control spark. Only reason I went to the HP (and nearly and probly "shouldha" gone MegaSquirt) is to get the "autotune" feature, as with no dyno time, and now one to ride shotgun and play with the laptop, I was about at the "max" of programming the 950
I also have two of those ECUs but I've got a fair amount of money into them.
I also have two of those ECUs but I've got a fair amount of money into them.
KitCarlson
Well-Known Member
Bill,
It never hurts to experiment with electronics.
I can help if you desire. I am most familiar AVR. While I have developed with, Intel, Zilog, Motorola, Ti including DSPs, ARM7, Atmel AVR is my favorite for task. An ATmega16 will easily do the interface tasks, or make a timing signal simulator. There are free evaluation C compilers, and low cost methods to load code. The requirements are not difficult, a few hours, and only several lines of code. There is a $20 USB ISP programming from pololu.com that loads code, has a 2 channel slow scope, and rx, tx converter for serial communication with micro. With it you will be able to measure timings in the development process.
It never hurts to experiment with electronics.
I can help if you desire. I am most familiar AVR. While I have developed with, Intel, Zilog, Motorola, Ti including DSPs, ARM7, Atmel AVR is my favorite for task. An ATmega16 will easily do the interface tasks, or make a timing signal simulator. There are free evaluation C compilers, and low cost methods to load code. The requirements are not difficult, a few hours, and only several lines of code. There is a $20 USB ISP programming from pololu.com that loads code, has a 2 channel slow scope, and rx, tx converter for serial communication with micro. With it you will be able to measure timings in the development process.
KitCarlson
Well-Known Member
I have been mostly absent due to my kitcar project. It does use my EMS. It installed easily, starts and runs well. I need to do more tuning, but need a revised title and plates. The good part, the Barracuda EFI, is one less deep in the project que.
My free website from ATT has been terminated. Here is a new one have just spent a few minutes on. https://sites.google.com/site/kitcarlsonems/
My free website from ATT has been terminated. Here is a new one have just spent a few minutes on. https://sites.google.com/site/kitcarlsonems/
KitCarlson
Well-Known Member
Thanks Del. I am trying to finish projects before I die, it is hard because at the same time they kill me. My hands are shot, taking a few days off to heal. With the changes, in weather, they are a bloody mess.
BillGrissom
Well-Known Member
Oops
BillGrissom
Well-Known Member
Kit, I feel your pain. My early A's are just sitting in the garage, as my wife notes almost daily. I did fix my wife's car today, but she won't remember that long. Currently, getting my 85 Mercedes ready for a son, and pulled a shoulder muscle in changing a ball joint. Before that, re-laying the tile roof and extending a driveway teamed up to pull a butt muscle (still feel it). I hope I get to my EFI plans eventually, but something else is sure to break before then. I can only dream of the time I would have if I can afford to retire in 10 years.
KitCarlson
Well-Known Member
I am much better now, pulling a few fiberglass slivers, superlue on the corners of my thumbs, and vitamin e cream.
I am struggling with the concept that things do not last forever. I had the kitcar 90% assembled in 1992. Then it got burried in the shop. I though only a few weeks to complete turned to a few months, with more work to go. I discovered the new 21 year old 4 wheel disk brakes are now locking up. I am gaining, and see completion of projects possible. I plan to find an interesting project, and do it in a more organized work area.
About retirement. Stopping the day job does not often yield more time. Our children often have things for us to do. I have been spending time with my youngest 33 year old son. I am trying to help him to work on house remodling and automotive. He is about where I was at half his age. He is learning, so that is good. Both my sons are engineers, but not so hands on. I think one has to be born with it.
I am struggling with the concept that things do not last forever. I had the kitcar 90% assembled in 1992. Then it got burried in the shop. I though only a few weeks to complete turned to a few months, with more work to go. I discovered the new 21 year old 4 wheel disk brakes are now locking up. I am gaining, and see completion of projects possible. I plan to find an interesting project, and do it in a more organized work area.
About retirement. Stopping the day job does not often yield more time. Our children often have things for us to do. I have been spending time with my youngest 33 year old son. I am trying to help him to work on house remodling and automotive. He is about where I was at half his age. He is learning, so that is good. Both my sons are engineers, but not so hands on. I think one has to be born with it.
-
